<- file 96regrII.html ->

... error-in-variables? Type II regr. (refs). Lorenz.

=====================Dave Lorenz, 24 Jan 1996========ssm,sse From: lorenz@nwq1dmnspl.cr.usgs.gov (Dave Lorenz) Subject: Re: Significance tests in Model II regression Message-ID: <1996Jan24.155806.7307@rsg1.er.usgs.gov> In article <4e3c5p$fn2@news.ccit.arizona.edu>, sfm@manduca.neurobio.arizona.edu (Stephen Matheson) writes: > I have a pretty standard biological problem in which I have > measured two variables which I suspect have a functional > relationship. Both are subject to error and thus model I > regression is not appropriate. Based on recommendations > given by Sokal & Rohlf (_Biometry_, 3rd edition, > W.H. Freeman & Co., 1995, pp. 541-5), I have concluded that > regression analysis should proceed via the reduced major > axis method. > > I am confused about how I might proceed with a comparison > of two different regression lines. My experiment involves > comparing the relationship between the aforementioned > variables under two treatment conditions. In other words, > I wish to test the hypothesis that the treatment alters > the relationship between the two variables. Sokal & Rohlf > provide examples of significance tests between regression > lines (pp. 493-9), but I can't tell whether or how to apply > these methods to my Model II situation. > > I look forward to suggestions and comments, perhaps in the > form of reading assignments :-). > -- > Steve Matheson sfm@neurobio.arizona.edu > "...lips that speak knowledge are a rare jewel." --Proverbs I am aware of three appraoches to determining the significance of model II regressions. Fortran code is available for each of these three approaches. I. Authors: G. Fasano and R. Vio, Astronomical Observatory of Padua, Padua, Italy. Reference: Newsletter of Working Group for Modern Astronomical Methodology, Issue 7, Sept. 1988, pp. 2-7. (Newsletter edited and distributed by F. Murtagh and A. Heck.) II. AUTHOR: F. Murtagh, ST-ECF, Garching. Version 1.0 9-July-1986 (Following discussions with J. Melnick and L. Lucy.) Reference: D. York, "Least squares fitting of a straight line", Canadian Journal of Physics, 44, 1079-1086, 1966. See also: M. Lybanon, "A better least squares method when both variables have uncertainties", Americal Journal of Physics, 52, 22-26, 1984. and: F.S. Acton, "Analysis of Straight Line Data", Dover, New York, 1959, Chapter 5. III. Author: B.D. RIPLEY 1981,1986 Reference: B.D. Ripley and M. Thompson, Regression techniques for the detection of analytical bias, Analyst, 112, 377-383, 1987. -- Dave Lorenz (lorenz@usgs.gov) =================Hans-Peter Piepho, 26 Jan 1996========ssc Message-ID: <9601261357.AA16381@fserv.wiz.uni-kassel.de> From: Hans-Peter Piepho <piepho@WIZ.UNI-KASSEL.DE> Subject: Re: regression when x has error >For estimting a and b in > y=a*x+b, >the traditional linear regression method is biased if there is >observation error in the independent variable x. > >Is there any better method for this situation? > > See Sokal and Rohlf. 1995. Biometry. 3rd ed. Freeman. under Model II regression ___ Hans-Peter Piepho * * * * * * * * * * * * * * * * * * * * * * * * * * * * *

Document by Rich Ulrich. E-mail to wpilib+@pitt.edu

FAQ top.

Ulrich home page.

Ulrich FAQ. http://www.pitt.edu/~wpilib/stats99.html